Metal oxide silicon field effect transistors ("MOSFETs") are in common use in integrated circuits performing a variety of functions such as microprocessors, memory devices, etc. A typical MOSFET 10 is illustrated in FIG. 1. The MOSFET 10 is formed on a substrate 12 of a silicon wafer which has been doped to be of a particular type which, in the example of FIG. 1, is p-type. An active portion of the MOSFET 10 is formed between areas 16, 18 of a field oxide layer of silicon dioxide SiO.sub.2. The field oxide areas 16, 18 are typically formed by exposing the silicon wafer 12 to oxygen at an elevated temperature thereby allowing the oxygen to react with the silicon. The field oxide also forms a gate oxide layer 20.
The substrate 12 on opposite sides of the gate insulator 20 is doped with a type different from the doping of the substrate 12. In the example of FIG. 1, n-type regions 30, 32 are formed in the p-type substrate 12. Deeper n-type regions 34, 36 are also formed in the center of the n-type regions 30, 32, respectively, for reasons that will be explained below. The region 30 forms the source of the MOSFET 10 while the region 32 forms the drain of the MOSFET 10. A gate 40 of the MOSFET 10 is formed on the gate oxide layer 20 by a conductive polysilicon layer.
After the above-described components of the MOSFET 10 have been formed, the substrate 12 is covered with a passivation layer 44, such as boron phosphorus silicate glass. Vias 48 formed in the passivation layer 44 provide a path for conductors 50, 52 to extend to the source region 30 and the drain region 32. Other conductors (not shown in FIG. 1) which are also part of a metalization layer extend to the gate 40 and to other MOSFETs and other components on the integrated circuit.
As mentioned above, relatively deep n-type regions 34, 36 are formed at the center of the source and drain regions 30, 32, respectively. The purpose of these deep n-type regions 34, 36 is to isolate the conductors 50, 52 from the substrate 12 in the event that the conductors 50, 52 penetrate too deeply into the source and drain regions 30, 32, respectively. However, in FIG. 1, the conductors 50, 52 are shown extending through vias in the passivation layer 44 and terminating at the surface of the substrate 12. In practice, the conductor 50, 52 may penetrate a sufficient distance into the regions 30, 32 to short to the substrate 12.
In operation, current flows from the drain conductor 52 to the source conductor 50 whenever the voltage applied to the gate 40 is greater than the voltage on the source conductor 50 by the threshold voltage V.sub.T of the MOSFET 10. Current flowing from the drain conductor 52 spreads out as it flows through the drain region 32 as illustrated by the arrows in FIG. 1. Similarly, current converges from several directions as it flows through the source region 30 to the source conductor 50.
As is well known in the art, the MOSFET structure shown in FIG. 1 is replete with capacitances. For example, the gate 40 and the substrate 12 form two plates of a capacitor separated by the gate oxide layer 20, and capacitances are also formed between the gate 40 and the n-type regions 30, 32 as well as between the n-type regions 30, 32 and other components. The switching speed of a MOSFET is a function of the time constant of the various components. The time constant is equal to the product of resistance and capacitance. Thus, for example, a higher resistance in the source region 30 results in a longer time for the source-to-drain voltage to reach a desired magnitude. An important factor in this high resistance is the resistance along the surface layer of the n-type regions 30, 32 as the current flows to or from the conductors 50, 52 as shown by the arrows, as explained above. If the conductors 50, 52 could contact the entire surfaces of the n-type regions 30, 32, respectively, or the surface resistances of the regions 30, 32 could be reduced, the switching time constants of the MOSFET 10 could be correspondingly reduced. A faster time constant would allow microprocessors, memory devices, etc., composed of MOSFETs to operate at a higher speed.
One conventional technique for reducing the resistance of MOSFETs is illustrated in FIG. 2 in which components of the MOSFET 60 that are identical to the MOSFET 10 of FIG. 1 have been provided with the same reference numeral. The prior art MOSFET 60 uses a siliciding process explained below with reference to FIG. 2. In the initial process steps, the field oxide areas 16, 18, the gate oxide layer 20, the gate 40 and the n-type regions 30, 32, 34, 36 are formed in the substrate 12, as illustrated in FIG. 2A.
The MOSFET 60 differs from the MOSFET 10 of FIG. 1 by the addition of silicide layers 64, 66 to the surface of the n-type source and drain regions 30, 32, respectively. Basically, siliciding is a process by which a metal silicon compound is formed in the surface of a silicon substrate 12 to reduce the surface resistance of the substrate 12. Siliciding is typically performed by coating the surface of the substrate 12 with a metal such as titanium, tungsten or molybdenum. The metal is then allowed to react with the silicon to form titanium silicide, tungsten silicide or moly silicide, respectively. Since siliciding consumes a portion of the surface of the silicon, the silicide layers 64, 66 shown in FIG. 2B extend below the original surface of the substrate 12. As shown in FIG. 2B, the silicide layers are masked by the field oxide areas 16, 18 in a process known as self-aligned siliciding, also known as saliciding. However, the silicide layers 64, 66 may also be formed using a dedicated mask (not shown).
After the silicide layers 64, 66 have been added as shown in FIG. 2B, the passivation layer 44 and the metal conductors 50, 52 are added, as shown in FIG. 2C. As further shown in FIG. 2C, the relatively low surface resistance of the silicide layers 64, 66 allows the current to spread out and flow directly into the drain region 32 and out of the source region 30. This reduced resistance allows spurious capacitances in the MOSFET 60 to be charged more rapidly. As a result, microprocessors, memory devices, and other semiconductor devices fabricated using a siliciding process as shown in FIG. 2 can operate at relatively high speeds.
Siliciding is also used to reduce the surface resistance of active areas of bipolar junction transistors 70, as illustrated in FIG. 3. As shown in FIG. 3A, a bipolar junction transistor 70 is fabricated by first forming a field oxide layer on a silicon substrate 72 at areas 74, 76, 78. As shown in FIG. 3A, the substrate 72 is doped with p-type material to form a p-type substrate 72. Next, as shown in FIG. 3B, the exposed areas of the substrate 72 between the field oxide areas 74-78 are doped to form an n-type collector 80 and an n-type emitter 82. As with the MOSFET 60 shown in FIG. 2, relatively deep n-type regions 84, 86 are also formed to prevent conductors from shorting to the substrate 72, as explained below. The n-type regions 80, 82 and the p-type substrate 72 form an N-P-N transistor as shown schematically at 88.
In the next step, silicide layers 90, 92 are formed on the exposed surfaces of the collector and emitter regions 80, 82, respectively. As with the MOSFET 60 of FIG. 2, the silicide layers 90, 92 are applied using the field oxide areas 74, 76, 78 as a mask in a saliciding process. However, a dedicated mask may also be used. After the silicide layers 90, 92 have been applied, a passivation layer 96 of a suitable material, such as boron phosphorous silicate glass, having etched vias 98 is deposited over the substrate 72, as illustrated in FIG. 3D. Finally, conductors 100, 102 are formed in a metalization layer as shown in FIG. 3E. The conductors 100, 102 extend through the vias 98 to contact the silicide layers 90, 92, respectively. The silicide layers 90, 92 allow the current to easily spread out and flow directly through the n-type regions 80, 82, as illustrated by the arrows in FIG. 3E. By reducing the surface resistance of the n-type regions 80, 82, the silicide layers 90, 92 allow the transistor 70 to switch more rapidly. As a result, semiconductor devices formed with the transistor 70 shown in FIG. 3 can operate at a higher speed.
Although the siliciding processes shown in FIGS. 2 and 3 have improved the operating speed of MOSFET and bipolar junction transistors, the siliciding process has caused certain problems. Specifically, the silicide layers can reduce the breakdown voltage between a region doped with one polarity and a substrate doped with the other polarity. This problem is illustrated in FIG. 4. Because of the manner in which the field oxide areas 16, 74 forms, it has an exposed sloped edge 110 and a burried sloped edge 112 formed in the substrate. The exposed sloped edge 110 joins the burried sloped edge 112 at an area known as the "bird's beak" 114. Similarly, because of the manner in which the doping of the n-type regions 30, 80 occurs, the n-type regions have a sloped boundary 116 formed in the substrate 12, 72. The structures shown in FIG. 4 makes the n-p junction at the bird's beak 114 relatively easy to break down, particularly since the silicide layer 64, 90 consumes the upper portion of the substrate 12, 72. As a result, the n-p junction between the n-type region 30, 80 and the substrate 12, 72 can easily break down at the bird's beak 14 at a relatively low voltage.
Although the problem of low breakdown voltages can exist in any MOSFET or bipolar transistor, it is primarily a problem with high voltage MOSFETs and bipolar junction transistors used for electrostatic discharge protection of MOSFET circuits. With reference to FIG. 5, a bipolar junction transistor 70 is commonly used to protect semiconductor circuits 130 in an integrated circuit 132 formed on a substrate. The semiconductor circuits 130 are connected to external circuitry through a connection pad 134. As is well known in the art, the semiconductor circuits 130 are composed of MOSFETs fabricated on a substrate of the integrated circuit 132. The bipolar junction transistor 70 protects the more fragile MOSFETs used in the semiconductor circuitry 130 by connecting the collector of the transistor 70 to the pad 134 and grounding the emitter of the transistor 70. The base of the transistor is connected to the substrate 72 of the integrated circuit 132, as illustrated in FIG. 3. It is important that the n-type regions 80, 82 of the transistor 70 do not break down at excessively low voltages. However, the n-p junctions of the transistor 70 may break down at excessively low voltages because of the presence of the silicide layers 90, 92 at the bird's beak 114, as shown in FIG. 4.
As a result of the above-described problems, semiconductor device designers must now generally choose between using siliciding to operate at a high speed while suffering the attendant reduced junction breakdown voltages or, in order to obtain an adequate junction breakdown voltage, fabricate semiconductor devices that are incapable of operating at a relatively high speed. There is therefore a need for semiconductor devices and a method of fabricating semiconductor devices that allows both a high breakdown junction voltage and low surface resistance to obtain a high operating speed.